Educational Theory [1 ed.] 9781620817834, 9781613245804

Citation preview

Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.

Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.

EDUCATION IN A COMPETITIVE AND GLOBALIZING WORLD

Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.

EDUCATIONAL THEORY

No part of this digital document may be reproduced, stored in a retrieval system or transmitted in any form or by any means. The publisher has taken reasonable care in the preparation of this digital document, but makes no expressed or implied warranty of any kind and assumes no responsibility for any errors or omissions. No liability is assumed for incidental or consequential damages in connection with or arising out of information contained herein. This digital document is sold with the clear understanding that the publisher is not engaged in rendering legal, medical or any other professional services.

EDUCATION IN A COMPETITIVE AND GLOBALIZING WORLD Additional books in this series can be found on Nova’s website under the Series tab.

Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.

Additional E-books in this series can be found on Nova’s website under the E-books tab.

EDUCATION IN A COMPETITIVE AND GLOBALIZING WORLD

EDUCATIONAL THEORY

JALEH HASSASKHAH

Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.

EDITOR

Nova Science Publishers, Inc. New York

Copyright © 2011 by Nova Science Publishers, Inc. All rights reserved. No part of this book may be reproduced, stored in a retrieval system or transmitted in any form or by any means: electronic, electrostatic, magnetic, tape, mechanical photocopying, recording or otherwise without the written permission of the Publisher. For permission to use material from this book please contact us: Telephone 631-231-7269; Fax 631-231-8175 Web Site: http://www.novapublishers.com NOTICE TO THE READER The Publisher has taken reasonable care in the preparation of this book, but makes no expressed or implied warranty of any kind and assumes no responsibility for any errors or omissions. No liability is assumed for incidental or consequential damages in connection with or arising out of information contained in this book. The Publisher shall not be liable for any special, consequential, or exemplary damages resulting, in whole or in part, from the readers’ use of, or reliance upon, this material. Any parts of this book based on government reports are so indicated and copyright is claimed for those parts to the extent applicable to compilations of such works.

Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.

Independent verification should be sought for any data, advice or recommendations contained in this book. In addition, no responsibility is assumed by the publisher for any injury and/or damage to persons or property arising from any methods, products, instructions, ideas or otherwise contained in this publication. This publication is designed to provide accurate and authoritative information with regard to the subject matter covered herein. It is sold with the clear understanding that the Publisher is not engaged in rendering legal or any other professional services. If legal or any other expert assistance is required, the services of a competent person should be sought. FROM A DECLARATION OF PARTICIPANTS JOINTLY ADOPTED BY A COMMITTEE OF THE AMERICAN BAR ASSOCIATION AND A COMMITTEE OF PUBLISHERS. Additional color graphics may be available in the e-book version of this book.

Library of Congress Cataloging-in-Publication Data Educational theory / editor, Jaleh Hassaskhah . p. cm. Includes index. ISBN:  (eBook)

1. Education--Philosophy. 2. Education and state. I. Hassaskhah, Jaleh. LB14.7.E3967 2011 370.1--dc22 2011013679

Published by Nova Science Publishers, Inc. † New York

CONTENTS vii 

Preface Chapter 1

Adolescent Educational Outcomes: Do Peer Networks Matter? Igor Ryabov and Franklin Goza 

Chapter 2

Constructivism as Educational Theory: Contingency in Learning and Optimally Guided Instruction Keith S. Taber 

39 

Bridging Educational Technologies and School Environment: Implementations and Findings from Research Studies Francesca Bertacchini, Lorella Gabriele and Assunta Tavernise 

63 

One Night in Night School to High School in Brazil: Some Impressions Heslley Machado Silva 

83 

Pre-Service Mathematics Teachers’ Perceptions about Mathematics Problems and the Nature of Problem Solving Fatma Kayan Fadlelmula and Erdinç Çakiroğlu 

97 

Chapter 3

Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.

Chapter 4

Chapter 5

Chapter 6

Synthesis of Learning in the Patchwork Text: Patchwork Assessment or Patched Work Assessment? Kelvin Tan and Chong Siew Fong 

Chapter 7

E-Maths Angel Garrido 

Chapter 8

Cognitive Diversity in Interdisciplinary Educational Theory Development Don Ambrose 

Index

1 

109  121 

129  143 

Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.

Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.

PREFACE Educational theory can refer to either speculative educational thought in general or to a theory of education as something that guides, explains, or describes educational practice. In this book, the authors present current research in the study of educational theory including the role of peer social capital as a predictor of adolescent academic outcomes; constructivism as educational theory; educational technology experiments and pre-service mathematics teachers perceptions about the nature of problem-solving. Chapter 1 - Little attention has been paid to the role of peer social capital in the school context, especially as a predictor of adolescent academic outcomes. The present study uses multilevel models and a nationally representative sample to address this issue. Results reveal that in addition to those factors typically associated with academic outcomes (e.g., school composition), peer networks also had a significant impact on educational achievement and attainment. As in many prior studies, school composition was a significant predictor of achievement for students of all racial/ethnic groups. Although academic attainment was worse in schools where low-income students were concentrated, for some racial/ethnic groups educational attainment increased with higher concentrations of minority students. Furthermore, peer social capital, measured as the average achievement of a peer network, was a significant predictor of both educational achievement and attainment for all racial/ethnic groups. In addition, and counter to some earlier studies, results revealed that segregated peer networks among African-Americans may lead to better academic achievement and attainment than school settings with more integrated friendship networks. Chapter 2 - Constructivism is a major referent in education, although it has been understood in various ways, including as a learning theory; a philosophical stance on human knowledge; and an approach to social enquiry. In terms of informing teaching, constructivism has variously been seen by different commentators as a basis for progressive, mainstream or failed approaches to pedagogy. This is unfortunate, as the different ways the term has been interpreted have confused debate about the potential of constructivism to contribute to planning effective teaching. This chapter sets out the basis of one version of constructivism: that which is informed by findings from both cognitive science, and from educational studies exploring learners’ thinking about curriculum topics and about classroom processes. A key concept here is the way in which new learning is contingent on features of the learner, the learning context and the teaching. This version of constructivism (which has been widely embraced) offers a theoretical basis for designing effective pedagogy that is accessible to classroom teachers.

Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.

viii

Jaleh Hassaskhah

The chapter will explain that although constructivism understood this way certainly offers the basis for learner-centred teaching, it is far from ‘minimally-guided’ instruction, as caricatured by some critics. Rather, a feature of this approach is that it does not adopt doctrinaire allegiance to particular levels of teacher input (as can be the case with teaching through discovery learning, or direct instruction) but rather the level of teacher guidance (a) is determined for particular learning activities by considering the learners and the material to be learnt; (b) shifts across sequences of teaching and learning episodes, and includes potential for highly structured guidance, as well as more exploratory activities. When understood in these terms, constructivism provides a sound theoretical basis for informing teaching at all levels, and in all disciplines. Chapter 3 - The aim of this chapter is to present educational technology experiments with school students, organized as laboratories. Among these laboratories, we introduce arrangement and findings of the Virtual Theatre and the Edutainment Robotics Laboratories, also presenting that based on the construction of Chua’s circuits. Regarding the first one, the objective has been the manipulation of virtual contents, as well as the measure of both acceptation of technology and learning. In particular, we have tested a Virtual Theatre software as an educational tool in grammar classes of different schools in Cosenza (Italy). The system has been presented and a list of tasks has been provided, including the manipulation of: - the content “script of the story” in a group writing laboratory; - sounds for the recording of dialogues; - virtual agent to model/animate on the basis of the script to perform; - the performance in the virtual theatre. As regards the Edutainment Robotics Laboratory, the objective has been the investigation of the cognitive strategies showed by students in building and programming a Lego MindStorms Robot, adopting a systematic methodology in data collection. The aim has been the leading of subjects in the acquisition of new knowledge as well as the development of advanced cognitive skills in problem solving, in thinking strategies and in the acquisition of new concepts. The Robotics Laboratory has consisted in two parts: the first phase has foreseen some key theoretical lessons about Robotics and the programming language; in the second phase, the task “to build and program a robot able to cross an arena in order to take part in a race” has been assigned to each group. At the end of the educational activities, subjects have tested the robot behaviour and presented a project report (a documentation on each phase of the work). Reports have showed the work strategies, the modalities of problem solution, as well as the cognitive strategies adopted by students in programming the robot. Regarding the laboratory devoted to the construction of Chua’s circuits, its aim has been the dissemination of new science concepts among young students during the project “Chaos at School: building a Chua’s circuit, simulating its behaviour and using it to create sound and music”. The purpose of the project was to familiarize high school students with chaos and complexity related concepts, motivating them to build the physical circuit, simulate its behaviour and then to create patterns and music by software applications. Results have showed that the direct contact with the circuit has encouraged students to think, to formulate hypotheses and to test their hypothesis through experiments. Hence, results of all experiences have shown that an active learning can remarkably enhance students’ learning efficiency; moreover, a rich interaction can provide fruitful feelings of participation in the educational process. In this view, the tension between new forms of learning and old forms of schooling could not be solved with the victory of one on the other, but through a bridge between the two.

Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.

Preface

ix

Chapter 4 - Introduction: This study aims to follow a typical night of high school night in a public school, trying to demonstrate its characteristics from the perspective of students and especially from the classroom. Intends to analyze to the relevant literature. Method: We used the observation as the main instrument of research. The issues findings were used in guided interviews with students, intending to develop a qualitative analysis. Results and Conclusions: Students, counselors and teachers resent the educational policies that sought to reduce repetition and suit the age, such as accelerated classes. The students revealed a desire by teachers who meet with their proposals, which are severe and prepare their lessons, although occasionally give signals that take advantage of the loopholes of the lack of limit to have fun. It is necessary that the high school to review their roles and allow the issues revealed by his young students, such as the necessity for compliance with basic objectives can be analyzed and answered promptly. Chapter 5 - Problem solving is an important component of mathematics education, mainly because it provides an environment for students to reflect on their conceptions about the nature of mathematics, and develop a relational mathematical understanding. Due to its powerful characteristics, problem solving has been given value in mathematics education as a skill to be taught, as a goal for mental development, and as a method for teaching. Especially, for the last three decades, there have been attempts all around the world to make problem solving the focus of school mathematics rather than being an isolated part of mathematics curriculum. This study aimed to investigate pre-service elementary mathematics teachers’ beliefs about mathematics problems and the nature of problem solving. The sample of the study consisted of 244 senior undergraduate students studying in Elementary Mathematics Teacher Education programs at 5 different universities in Turkey. The data were collected during the spring semester of 2005-2006 academic years. Participants completed a survey composed of three parts as demographic information sheet, questionnaire items, and nonroutine mathematics problems. The results of the study indicated that, in general, pre-service teachers held positive beliefs about mathematical problem solving. However, a number of pre-service teachers held several traditional beliefs about problem solving, such as following predetermined sequence of steps while solving mathematics problems and the importance of computational skills in mathematics education. In addition, it was found that a number of preservice teachers did not value problems that do not cover any topic in the curriculum, problems that do not involve any number and problems that take a long time to solve. Chapter 6 - The Patchwork Text is a recent assessment innovation that seeks to enhance students’ synthesis of learning over a period of time through engagement with progressive tasks culminating in a final synthesized assignment. However, this is not dissimilar to a form of plagiarism known as ‘Patchwriting’, wherein students unfamiliar with academic convention resort to paraphrasing pieces of published text and representing the consequent miscellany of text as their own. It is argued that students may likewise resort to Patchwriting in Patchwork Text assessment. In this article, the Structure of Observed Learning Outcomes (‘SOLO’) taxonomy is used to explain how certain forms of Patchwork text design may lead to Patchwriting, and suggests how Patchwork text assessment may be designed to minimize such risks. An example of minimizing Patchwriting is provided in the form of a Patchwork Text assessment for English Literature. Chapter 7 - Some reflections which proceeds from teaching and research on the areas of Mathematics and Computation, which have taken the author to conclude that both fields lose

x

Jaleh Hassaskhah

Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.

perspective and potential when they are kept disconnected, and only they have felt when they work coordinated, looking each of them for the natural support in other one. Chapter 8 - Arguments over the nature and purpose of theory have abounded within and beyond educational fields. One counterproductive scenario is to leave education atheoretical and conceptually unguided. The opposite counterproductive scenario is to have education dominated by a grand hegemonic but inadequate, narrow theory. The phenomena of interest in education are highly varied and complex; consequently, educational fields require guidance from multiple theoretical perspectives, which can shed light on specific phenomena. Only an interdisciplinary search covering multiple levels of analysis can bring into play an adequate array of theories that can illuminate the intricate, complex, multiple dimensions of learners, educators, classrooms, school systems, and the sociopolitical, ideological, and economic contexts that influence them. Employing the concept of cognitive diversity, this chapter describes the conceptual territory from which useful theory can be drawn. This territory includes levels of disciplinary scope from macro-societal to immediate contextual, to the level of the individual, to the organic-biological, to the cellular biological, to the molecular-genetic, and possibly even to the subatomic. Examples of theoretical perspectives from these various levels are provided, along with some ways in which they can guide the work of educational researchers and practitioners.

In: Educational Theory Editor: Jaleh Hassaskhah, pp. 1-38

ISBN 978-1-61324-580-4 © 2011 Nova Science Publishers, Inc.

Chapter 1

ADOLESCENT EDUCATIONAL OUTCOMES: DO PEER NETWORKS MATTER? Igor Ryabov and Franklin Goza Bowling Green State University, Bowling Green, Ohio, USA

Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.

ABSTRACT Little attention has been paid to the role of peer social capital in the school context, especially as a predictor of adolescent academic outcomes. The present study uses multilevel models and a nationally representative sample to address this issue. Results reveal that in addition to those factors typically associated with academic outcomes (e.g., school composition), peer networks also had a significant impact on educational achievement and attainment. As in many prior studies, school composition was a significant predictor of achievement for students of all racial/ethnic groups. Although academic attainment was worse in schools where low-income students were concentrated, for some racial/ethnic groups educational attainment increased with higher concentrations of minority students. Furthermore, peer social capital, measured as the average achievement of a peer network, was a significant predictor of both educational achievement and attainment for all racial/ethnic groups. In addition, and counter to some earlier studies, results revealed that segregated peer networks among African-Americans may lead to better academic achievement and attainment than school settings with more integrated friendship networks.

The United States continues to become more ethnically and racially diverse (U.S. Census Bureau 2000) and this is especially so among the school-age population (Bean and Stevens 2003; Hirschman 2001; Orfield, Eaton, and Jones 1997). Between 1991 and 2001 the white share of public school enrollment fell from 67.4% to 60.3%, while the percentages of African-American, Latino, Asian, and Native American youth all increased (U.S. Department of Education 2002). Paradoxically, there was a simultaneous increase in school racial segregation for the first time since the 1955 Brown v. Topeka decision. Harvard’s desegregation project found that the percentage of black students attending high-percent

Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.

2

Igor Ryabov and Franklin Goza

minority schools fell from 76.6% in 1968-69 to 62.9% in 1980-81, but by 1996-97 had increased to 68.8% (Orfield and Yun 1999). Furthermore, by 2003 the enrollment of black students in predominantly white schools was lower than in any year since 1968 (Frankenberg, Lee, and Orfield 2003). In fact, today many inner city schools are more racially segregated than they were in 1955 (Orfield, Eaton, and Jones 1997). However, today the most segregated minority group is Latinos, not blacks, and their segregation levels have steadily increased over the past thirty years (Orfield and Yun 1999). Despite the fact that many schools are racially and/or ethnically segregated, their student bodies remain extremely diverse (Clotfelter 2001). According to Bankston and Caldas (1998: 534), although “segregated schools are not and have never been the products of selfsegregation by minority group members” the vast majority of teens is homophilic and prefers in-group associations (Joyner and Kao 2000; Kubitschek and Hallinan 1998; Moody 2001). Even when schools have diverse populations, students may not be integrated to the extent that members of distinct ethnic groups regularly interact with one another (Cohen 1975; Epstein 1985; Maran 2000; Tatum 1999). Thus even relatively diverse student populations do not ensure high levels of interracial contact among students. Numerous studies have examined the relationship between school social composition and educational outcomes. However, prior research has not paid enough attention to the possible effects of school-based friendships and networks on educational outcomes. Thus the question of whether school-based racial and/or ethnic social capital can explain adolescent educational outcomes remains unanswered. This research will analyze the importance of various schoollevel factors on student academic achievement while controlling for other family- and individual-level measures. The school-level predictors to be examined include peer network segregation, school socioeconomic status, and school racial/ethnic composition. These enable us to address the following questions: (1) Do youth with more racially and/or ethnically segregated peer networks have worse academic achievement than those with less segregated peer networks?; (2) Do youth in schools with higher percentages of minority enrollment have lower academic achievement than those in schools with smaller percentages of minority enrollment?; (3) Do youth in high socioeconomic status (SES) schools have better academic achievement than those in low SES schools?; and (4) How do the results for the first three questions vary across racial/ethnic groups? This study is both conceptually and empirically significant. First, it updates and expands the implications of earlier research by incorporating more recent theoretical developments, empirical findings, and statistical techniques (Bryk and Raudenbush 1992; Raudebusch and Willms 1995). By so doing, we capture the essence of the school context by viewing adolescent networks and school composition as intertwined rather than isolated from one another. Second, we examine school-level effects differentially by race and ethnicity. As this is done we explore the tenets of oppositional culture theory, which posits that the orientation of friendship ties towards co-ethnic and co-racial peers hurts minority achievement (Ogbu 1981). Third, the present study draws on nationally representative data which are analyzed using multilevel modeling techniques. Fourth, school-level characteristics are examined while controlling for measures of family background and family social capital. Fifth, the present study examines academic performance both cross-sectionally and longitudinally. Sixth and most importantly, we examine both race and class, two key “family background” components, as individual- and school-level predictors of academic achievement.

Adolescent Educational Outcomes: Do Peer Networks Matter?

3

This chapter proceeds by first reviewing relevant literature on school segregation and peer networks. Next is a discussion of the data and methods used as we also describe the individual- and school-level independent measures to be analyzed and the hypotheses examined. This is followed by our results and conclusions.

Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.

RELEVANT RESEARCH ON SCHOOL SEGREGATION Although most scholars agree that education is an important mechanism for social mobility, some also believe that schools reproduce social inequality (Bankston and Caldas 2002; Carnevale 1999; Kahlenberg 1996; Roscigno 1998). The influences of race and social class extend well beyond the family realm, as they shape school attendance patterns and contribute to the creation of highly segregated school contexts, all of which affect academic achievement. Many of the mechanisms regarding how segregation affects achievement remain unknown. One intuitive explanation often advanced is that because school racial composition determines one’s ability to make friends with students from other racial and ethnic groups, an integrated school would provide the best possibility for frequent interracial contact and interracial information transfer (Bankston and Caldas 1996; Coleman et al. 1966; Longshore and Prager 1985; Mahard and Crain 1983; Schofield 1993). Thus the importance of interracial peer contact in schools has been of particular interest to social scientists. It has also served as one rationale for pursuing school desegregation. In the 20th century the political issues surrounding racial/ethnic integration, generally, and school desegregation in particular, aroused intense debate as numerous social scientists attempted to document how school (de)segregation affects the academic achievement of both minority and majority students. Thus far the evidence on the effects of desegregation has been mixed. Some studies observed short-term positive effects of desegregation on the math and verbal scores of black students (e.g., Hoxby 2000; Schofield 1993), while others found little or no evidence linking racial segregation to academic achievement (Armor 1995; Ascher 1992; Crain and Mahard 1978; Jencks 1972; Leake and Leake 1992; Rivkin 2000). Efforts to synthesize research findings on the effects of desegregation have led some to conclude that the evidence is so mixed or contradictory that reliable conclusions are impossible (e.g., Bankston and Caldas 2002). One explanation for the apparent ambiguity of much of this research is that the effects of desegregation vary enormously from community to community and from school to school. Others suggest that the best indicator of school quality is not level of integration, but rather socioeconomic composition (e.g., Kahlenberg 1996, 2001). Much evidence shows that high-poverty schools reduce the educational performance of children, even when controlling for children’s own class and race (e.g., Bankston and Caldas 1996; Entwisle and Alexander 1992). As such, the collective evidence accumulated by desegregation research made Orfield (1978:78) conclude: “Educational research suggests the basic damage inflicted by segregated education comes not from racial isolation but from the concentration of children from poor families.” The most influential study to date, the Coleman Report, found that the “beneficial effect of a student body with a high proportion of white students comes not from racial composition per se but from the better educational background and higher educational aspirations that are, on the average, found among whites” (Coleman et al. 1966:307). Using

Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.

4

Igor Ryabov and Franklin Goza

data from over 600,000 students and teachers across the country, the report found that educational outcomes were primarily influenced by individual factors, such as a student’s adaptation to school and the student’s family background. Although the Report noted that individual factors supersede school-level factors, it also confirmed that low-income students experience greater achievement gains when they attend middle-class schools than when they attend high-poverty schools. The Report further found that the socioeconomic composition of a school’s student body is more highly related to achievement, independent of students’ own social background, than any other school factor. Accordingly, both poor blacks and whites should benefit from attending a middle-class black school, whereas poor blacks would not enhance their academic achievement by attending schools largely populated by poor whites. A number of criticisms that may be raised regarding both the Equality of Educational Opportunity (EEO) data used in the Report as well as the methodological approach undertaken (see Jencks 1972 and Madaus et al. 1979). For instance, most of the school-level analyses were reduced to analyses of correlation-covariance matrices where the order of inserting variables often determined the magnitude of correlation coefficients. Another important criticism of the Coleman Report is that, although it had access to student scores on standardized tests and grades, it used verbal ability as the primary dependent measure (Madaus et al. 1979). Perhaps it was the media’s oversimplification of the Report’s findings that lead to the controversy about it and the belief among some that schools do not make a difference. Indeed, the methodological limitations discussed above may have lead to an underestimation of the school effect. Numerous studies conducted after the Coleman Report concurred that the social class of a student’s classmates matters more than their race. In 1972, using the same EEO data, Jencks repeated the Coleman analyses and found that poor sixth-graders, regardless of race, attending a high-poverty school were academically years behind their poor peers who attended a middle-class school. Jencks (1972), like Coleman, did not find significant racial differences in this regard. Jencks’ study, however, is subject to the same criticism as the Coleman Report, as he also used a number of standardized tests as measures of achievement. Later studies would revisit these research questions with more complex statistical techniques.1 Among these studies, those that employed multilevel modeling are of special attention as they tend to produce more accurate estimates of the school effect (Raudenbush and Bryk 2002; Raudenbush and Willms 1995). Bryk and Raudenbush (1992) pioneered the use of Hierarchical Linear Models (HLM) for the purpose of producing more accurate statistical inferences from complex multilevel data. Using HLM and a sample of 7,185 students from 160 schools, they estimated that between-school variance accounted for 18 percent of the total variance in student math tests. Moreover, almost 70% of total between-school variance was explained by a single factor, the mean school SES. We caution against emphasizing the significance of this finding as Bryk and Raudenbush (1992) did not control for the schools’ racial/ethnic composition, a factor which is typically found to be one of the most significant school-level predictors of educational achievement (Bankston and Caldas 2002). However, 1

For instance, Chubb and Moe (1990), using longitudinal data, found the average SES of a school’s student body to be strongly associated with gains in academic achievement among high school students. More recently, SuiChu and Williams (1996) examined factors influencing the math and reading scores of eighth-graders and concluded that the effect of a school’s SES was as strong as that of the family SES. Lastly, Puma et al. (1997:73) analyzed a nationally representative sample of schools and concluded that “the poverty level of the school is negatively related to standardized achievement scores.”

Adolescent Educational Outcomes: Do Peer Networks Matter?

5

Bryk and Raudenbush (1992) did control for school type (i.e., private/public), a factor sometimes found to be significant, which may partially explain their high estimate of the variance explained by average school SES. The fact that Bryk and Raudenbush (1992) did not include extensive individual-level controls may also account for their high estimate of between-school variation in math test scores (i.e., approximately 20% as compared to only 10% in the studies by Coleman and Jencks). Building on Raudenbush and Bryk’s (1992) study, we will use the same software (i.e., HLM), but a different analytical strategy that satisfies the call for a more elaborate methodological approach to monitor the school effect.

Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.

RELEVANT RESEARCH ON PEER NETWORKS School segregation affects students in ways that go far beyond the effects of class isolation. To begin, the peer networks students develop and the friendships they form are important consequences of the school they attend. Recent research (Haynie 2002) suggests there is much to learn about these networks as they are far more complicated that previously believed. Furthermore, relatively little is known about how peer networks affect educational outcomes. The unique data used in this chapter will enable us to examine two aspects of adolescent networks believed to be important determinants of academic behavior. The first is peer network social capital, an individual-level measure, while the second, peer network segregation, is determined at the school-level. Each is briefly discussed below. Peer group theory predicts that the prospects for adolescent school success will vary depending on the peer group with whom adolescents most often come into contact (Coleman et al. 1966; Hallinan and Sørensen 1985). The peer group is the context in which adolescents are exposed to others, including role models. It involves contemporaneous behavioral influences and is always reciprocal (Coleman 1988; Coleman et al. 1966; Schneider and Coleman 1993). Chubb and Moe (1990:109) consider peer friendships at school to be a critical link between families and schools because “through their peers, students are influenced by the families of other students in a school.” The acquaintances and communications between students foster social capital because they make possible network connections among sets of individuals (Hallinan and Sørensen 1985; Harris et al. 2002; Kubitschek and Hallinan 1998; Morgan and Sørensen 1999). However, this social capital may be used to promote either positive or negative outcomes (McCarthy and Hagan 1995). Thus although peer groups may provide members with the opportunity to form positive skills and relationships, including those that may make them academically successful, they may also transmit less desirable behaviors such as those that make adolescents more likely to fail academically (Berndt and Keefe 1995; Wentzel and Caldwell 1997) or to engage in delinquent behavior (Haynie 2002). Various theoretical ideas have been advanced to describe the positive and negative aspects of peers on academic achievement. For instance, Ogbu (1978, 1981) has used oppositional culture theory to describe a cultural pattern within African-American and Latino communities whereby peers disparage academic achievement because it is perceived as “selling out” or “acting white” (Fordham and Ogbu 1986; Ogbu 1991). Ogbu (1978) argues that minority students tend to develop a collective oppositional culture, a frame of reference that actively rejects mainstream behaviors and undermines academic achievement. In other

Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.

6

Igor Ryabov and Franklin Goza

words, children in this situation are often ostracized for conforming to the educational system. As a result, Steinberg et al. (1992) argue that minority students receive less support for achievement from their peers of the same ethnic background and consequently do not fare as well in school as non-Hispanic white students. Just as links have been established between negative peer influence and academic outcomes (Berndt and Keefe 1995; Berndt et al. 1990), similar linkages may be established between positive academic outcomes and peer influence (e.g., Epstein 1983). For example, Carter (2003) reported that while black and Latino students rejected certain styles of speech, dress, and music as “acting white,” they nonetheless valued behaviors conducive to academic success, such as studying hard, getting good grades, and making the honor roll. The present study will determine which, if any, of these possibilities best describes how peer networks affect academic outcomes. A related way of examining the potential positive impact of networks on adolescent academic achievement is to examine the effect of ethnic social capital on academic outcomes. Borjas (1992, 1995) locates ethnic social capital within the ethnic group and its networks. This notion of ethnic social capital has primarily been used in studies of immigrants and assimilation (e.g., Portes 1998; Portes and Rumbaut 2001). Borjas hypothesizes that immigrant and minority children may experience increased chances of economic success when they develop in social environments with greater amounts of ethnic social capital. Ethnic groups and networks provide intergenerational transmissions of social and human capital, norms regarding educational attainment, as well as educational and employment opportunities. Those ethnic groups that resist acculturation and maintain high levels of ethnic solidarity may provide better opportunities for younger generations through the creation and diffusion of ethnic social capital. Because minority youth are often disadvantaged regarding other forms of social and financial capital, ethnic social capital may be a beneficial form for the educational outcomes of these adolescents, especially given their occasional reliance on peer based social capital to compensate for the lack of family social capital (Lin 1990; Zhou and Bankston 1998). However, thus far the lack of appropriate data (Haynie 2002), as well as the typical view that co-racial and co-ethnic peer influences among adolescents were considered liabilities, have meant that this possibility has not yet been systematically examined. This chapter, however, will carefully analyze the effect of ethnic social capital within the school context to determine how it affects the educational achievement and attainment of minority youth. In a related vein, various measures of family social capital (described below) will also be considered in multivariate analyses. The second aspect of peer networks this study will examine is peer network segregation. According to Hallinan (1982), the racial/ethnic composition of a student body determines the probability of interracial friendship formation by influencing the composition of friendship pools from which students draw. Most researchers consider interracial friendships beneficial for the academic performance of minority students (Chubb and Moe 1990; Coleman et al. 1966; Hawley and Smylie 1988; Roscigno 1998). The Coleman Report explained the benefits of school integration as the transmission of values. More specifically, socially acceptable patterns of behavior were diffused from the more privileged racial group to the less privileged one through interracial contact (Coleman et al. 1966; Gerard 1988). Other scholars stressed the importance of information transfer, which is facilitated in integrated environments (e.g., Chubb and Moe 1990). Still others (e.g., Hawley and Smylie 1988) argue that interracial

Adolescent Educational Outcomes: Do Peer Networks Matter?

7

friendships provide minorities with access to resources, means of self-presentation, and patterns of communication acceptable to the majority. In essence, virtually all prior research extols the virtues of integration and interracial friendships and suggests that they lead to improved academic performance. However, the lack of adequate data made it impossible for prior studies to systematically examine this oft made assertion. Below we examine the effects of peer network segregation to determine how it is related to both adolescent educational achievement and attainment.

HYPOTHESES

Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.

Several of the hypotheses examined in this chapter have already been tested before using different data sets and techniques. However, what is unique about the present study is that these research questions are now examined together with still others that have never before been addressed. The primary hypotheses examined in this study may be stated as follows: 1. Like prior studies of segregation (e.g., Bankston and Caldas 2002; Coleman et al. 1966; Jencks 1972), we incorporate school racial and ethnic composition as a key independent variable and hypothesize that it will affect adolescent achievement and attainment. More precisely, we expect that attendance at a school with high minority enrollment will be negatively associated with academic achievement and attainment. 2. Based on earlier research (e.g., Entwisle and Alexander 1992; Kahlenberg 1996, 2001; Orfield 1978), we hypothesize a direct relationship between average school SES and the academic achievement of its students. More specifically, attendance at a high SES school is expected to be associated with high academic achievement and vice versa. 3. We hypothesize that differences in academic performance are influenced by friendship preferences, especially with respect to race and ethnicity. Based on Borjas’ (1992, 1995) notion of ethnic social capital, and contrary to oppositional culture theory (Ogbu 1978, 1981; Fordham and Ogbu 1986), we expect to see significant interactions between race/ethnicity and one’s orientation towards ethnic social capital, as proxied by the peer network segregation index. In other words, minority youth preferences for inter-group rather than intra-group ties may positively affect their academic achievement and attainment. The additive and multiplicative effects of this possibility are analyzed below for all racial/ethnic groups considered. 4. Following Coleman (1988), who suggested social capital manifests itself not only in the structure of social groups and networks but also in the quality of relationships and the amount of support they provide, we hypothesize that the amount of social capital present in networks is directly related to academic success. Since we proxy this amount as network achievement, we expect to find a positive association between network achievement and an individual student’s achievement and attainment. 5. Research suggests that family based social capital is one of the most important factors influencing adolescent educational success (Dornbusch et al. 1987; Israel, Beaulieu and Hartless 2001; Stevenson and Baker 1987). Various scholars (e.g., Coleman 1990; Teachman et al. 1996) have distinguished between the structural

8

Igor Ryabov and Franklin Goza (e.g., family structure and size) and relationship components (e.g., the quality of parent-child relationship) of family social capital. An increasing number of studies suggest that the presence or absence of a parent, as well as family size may affect adolescent educational outcomes (Bridge et al. 1979; Hetherington 1998; McLanahan and Sandefur 1994; Nelson et al. 2001; Schneider and Coleman 1993; Tienda and Angel 1982; Thomson et al. 1994). Similarly, adolescent academic performance has been linked to instrumental parental practices that stimulate and monitor children’s educational progress (Conger et al. 1997; Galambos et al. 2003; Israel et al. 2001; McNeal 2001; Muller 1995; Steinberg et al. 1992; Zhou and Bankston 1998). Therefore, we hypothesize that adolescents from non-traditional (i.e., single parent and non-parent) and large families will do less well academically than adolescents from two-parent nuclear families. Second, we expect that all types of family social capital (i.e., parental expectations, involvement and supervision) will have a positive effect on students’ academic achievement.

Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.

Data and Methods The data used to investigate the aforementioned hypotheses is the National Longitudinal Study of Adolescent Health (hereafter, Add Health). This nationally representative, schoolbased data set was collected in three waves, in 1994-95, 1996 and 2001-2002, respectively. A sample of 80 high schools and 52 middle (feeder) schools from the U.S. was selected with unequal probability of selection.2 During Wave I all students present in the 132 selected schools the day the self-administered questionnaire was conducted were surveyed (N=90,118). These data were supplemented with information supplied by an official at each of the surveyed schools. A subset of students was randomly selected from the 132 schools (N=20,745) for in-home interviews, as was a parent or parent-figure. With the exception of educational attainment, which comes from Wave III, we rely on the Wave I data as they provide the most complete information on all variables of interest. Those cases with missing values on educational attainment in Wave III were excluded. Applying this selection criterion reduced our final sample size to 19,117 students from 129 schools. Auxiliary univariate analyses revealed that this reduction did not affect the distribution of sample. These detailed data enable us to rigorously measure peer networks in ways that were not possible with earlier datasets (Haynie 2001, 2002). The Add Health data are distinguished by their hierarchical structure. This structure enables us to interpret the multifaceted nature of student achievement as a function of individual- (e.g., sex and age) and school-level factors (e.g., school racial and ethnic composition). For our analyses the Hierarchical Linear Models (HLM) statistical package will be used since it incorporates such factors more efficiently than ordinary least squares regression. HLM also takes into account the error structures present at 2

Initially, 80 high schools were selected from a sampling framework of 26,666. Of these, 52 were eligible and agreed to participate. The other 28 schools were replaced by randomly selected similar high schools that matched on eight criteria (school size, school type, level of urbanization, percent white, grade span, percent black, census region, census division). A total of 52 feeder or middle-schools that matched the attributes of the selected high-schools were used. As a result, the composition of middle and high schools in the Add Health sample are very similar (Chantala and Tabor 1999). Our analyses indicate that key school-level variables (e.g., school composition and peer network segregation) in middle schools are not significantly different from those of the high schools.

Adolescent Educational Outcomes: Do Peer Networks Matter?

9

each level (see Bryk and Raudenbush, 1992 or Raudenbush and Bryk, 2002) for more information on HLM).

Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.

Dependent Variables This study will estimate both short- and mid-term school effects on the educational progress of middle- and high-school students. Accordingly, this project’s dependent variables are educational achievement, measured as GPA, and attainment, measured as high school graduation. These two variables are theoretically distinct. Although attainment as conceptualized by Blau and Duncan (1967) and Sewell and Hauser (1975) is a function of family background, and, to some degree, intellectual abilities, achievement is also related to one’s ability to adapt to their educational context (Bridge et al. 1979; Lareau 1989). More importantly, the aforementioned classical studies (Blau and Duncan 1967; Sewell and Hauser 1975) document that educational achievement has a long-term effect on attainment, which, in turn, has a profound effect on one’s children and so forth. Therefore, while achievement may determine attainment, the opposite is not true. Because of these conceptual differences between achievement and attainment, distinct analyses will focus on each dependent variable. Achievement was computed based on the grades adolescents reported they earned the prior year in four subjects (English, math, science and social studies). These four responses, ranging from 1 (D or F) to 4 (A), were averaged across subjects and converted to a standard 4-point GPA. Although slightly inflated, self-reported grades are highly correlated with grades reported on official transcripts (Dornbusch et al. 1990). Table 1 shows that the average GPA in Wave I was about 2.8. The measure educational attainment is culled from the Wave III results, which helps to address concerns about causal ordering. At that time respondents, then young adults between the ages of 18 and 26, were asked about the highest grade of regular school they had completed. Their answers range from “6th grade” (the lowest score) to “5 or more years of graduate school” (the highest score). Note that this measure is cohortspecific and censored from above and below. In other words, attainment and age are inextricably linked. Hence, although the variable’s distribution is approximately normal, it is because of factors unrelated to attainment. Controlling for age alone does not eliminate the problem of cohort-specificity. Because of this and the fact that the median and mode for Wave III attainment approximates graduation from high school, we transformed the original Add Health measure into a dichotomous outcome variable with at least high school graduation equal to 1. Thus, in contrast to achievement, which monitors GPA, educational attainment provides information about an educational transition this cohort may have experienced.

INDEPENDENT VARIABLES Individual-Level Measures Peer Social Capital. In this study, we employ two measures of peer social capital. One, discussed below in the section School-Level Variables, captures the structural component of

10

Igor Ryabov and Franklin Goza

peer social capital. The other is an individual-level measure computed as the mean GPA of a student’s peer network (henceforth peer network GPA). This measure relates to the actual amount of peer social capital and, hypothetically, peer support available to an individual student. Pre-constructed by the Add Health, this variable does not account for unequal network sizes. In other words, in those cases where networks are relatively small, peer network GPA is likely to approximate an individual student’s GPA, thereby creating a source of collinearity with the individual student’s achievement. To eliminate this bias we transformed the Add Health peer network GPA measure according to the following formula:

Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.

New Network GPA =

UNGPA × NS − Individal Student ' s GPA , NS − 1

where UNGPA = untransformed network GPA and NS = network size. Other individual-level variables are examined in an attempt to control for personal and family factors that might impact academic achievement. These include race/ethnicity, gender, age and frequency of involvement in extracurricular activities. Race and ethnicity are determined based on student responses. From these responses we created a series of dichotomous race/ethnicity variables for the categories African-American, Asian, Latino, and non-Hispanic white.3 The latter serves as the reference category in these analyses. Gender is a dummy variable with male serving as the reference category. Age is measured in complete years at the time of interview. We monitor the variable involvement in extracurricular activities because high-achieving students typically spend more time engaged in learning activities both in and outside of school than do lower-achieving students (Blum and Reinhart 1997). In the Add Health extracurricular learning activities may include reading, writing, arts and crafts and other activities. Information on these activities comes from the question: “During the past week how many times did you do hobbies, such as collecting baseball cards, playing a musical instrument, reading, or doing arts and crafts?” Response categories range from 0 “not at all” to 3 “5 or more times.” Table 1 reveals that the sample’s sex ratio is balanced with approximately equal proportions of male and female students. The average age of Wave I respondents in the summer of 1995 was 15 years. Approximately 65% were non-Hispanic white, 16% AfricanAmerican, 14% Hispanic, and 5% Asian. The modal score for extracurricular activities was 1 meaning that this group participated in 1-2 extracurricular activities during the past week. Prior research shows that adolescent educational outcomes were associated with immigrant generational status (e.g., Kao and Tienda 1995, Orfield and Yun 1999). This is not surprising given that theories of immigrant adaptation have long predicted differential outcomes across immigrant generations (see Hirschman 2001 for more information about theories of immigrant assimilation). “Straight-line” assimilation posits continuous improvement with each successive generation while segmented assimilation theory (Portes and Zhou 1993) emphasizes contexts of reception. Because all assimilation theories underscore the significance of generational status, we created three dummy variables to monitor respondents’ generational status. 3

Asian and Latino adolescents were not divided into distinct ethnic origin groups (i.e., Chinese, Cubans, etc.) because these groups are too small to make statistically significant inferences about their educational achievement.

Adolescent Educational Outcomes: Do Peer Networks Matter?

11

Table 1. Descriptive Statistics of Study Variables (N=19,117) a

Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.

Weighted Mean Peer Social Capital Peer Network Segregation Index b Peer Network Achievement School Composition Average SES Percentage of Minority Students b Percentage of High-School Students Dependent Variables Educational Achievement Educational Attainment Race/Ethnicity African-American Asian Latino Non-Hispanic whites Family Structure Two-Parent Household Single-Parent Household Non-Parent Household Large Household SES Parents’ Education Family Income b Family Social Capital Parents’ Educational Expectations Parents’ Involvement Parents’ Supervision Individual-Level Controls Age Male Immigrant Generation 1 Immigrant Generation 2 Immigrant Generation 3 Extracurricular Activities a b

St. Minimum Maximum Deviation

0.25 2.81

0.10 0.69

0.48 1.00

0.75 4.19

0.85 0.26 0.62

0.02 0.18 0.48

0.96 0.00 0.00

0.65 0.69 1.00

2.82 0.86

0.76 0.35

0.77 0.00

4.17 1.00

0.16 0.05 0.14 0.65

0.36 0.23 0.35 0.45

0.00 0.00 0.00 0.00

1.00 1.00 1.00 1.00

0.59 0.24 0.17 0.20

0.50 0.43 0.38 0.40

0.00 0.00 0.00 0.00

1.00 1.00 1.00 1.00

6.84 5.27

2.11 1.48

0.00 0.26

10.85 14.21

4.33 0.43 3.83

0.89 0.29 0.70

1.00 0.02 1.00

6.58 1.82 5.67

14.98 0.49 0.05 0.10 0.85 1.39

1.66 0.01 0.21 0.30 0.36 1.56

11.00 0.00 0.00 0.00 0.00 0.00

21.00 1.00 1.00 1.00 1.00 22.00

All variables are from Wave I except for educational attainment, which is from Wave III. Percentage of minority students, peer network segregation index, and family income were transformed by the Box-Cox method in order to satisfy the multilevel normality condition of HLM (see more on HLM in Raudenbush and Bryk 2002).

Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.

12

Igor Ryabov and Franklin Goza

Foreign-born adolescents are coded as immigrant generation one. U.S.-born children with at least one foreign-born parent are distinguished as generation two and generation three is comprised of those born in the U.S. with two U.S.-born parents. Socioeconomic Status (SES). Household income and parents’ education are included in an attempt to control for family SES, a factor often linked to adolescent academic achievement (e.g., Bridge et al. 1979; Conger et al. 1997; Lareau 1989; McLoyd 1998). The income measure was obtained from the parental response to the question: “About how much total income, before taxes did your family receive in 1994? Include your own income, the income of everyone else in your household, and income from welfare benefits, dividends, and all other sources.” Responses are coded in units of 1000 and range from 0 to 999. Those cases with negative income were recoded as zeros because reports of negative household income, as opposed to individual income, may indicate debt and, thus, differ in nature from the income measure.4 The parental education measures came from items asking: “How far did she [mother] go in school?” or “How far did he [father] go in school?” This is a measure of the highest level of education completed. Response categories range from “eighth grade or less” (coded 1) to “graduate training beyond a four-year college or university” (coded 9). Parents’ education is recoded as the highest educational attainment of both parents. In order to account for family structure all family social capital measures, except parents’ education, were constructed as the average response for both parents, if available, and as simple measures if responses for only one parent were available. Multiple imputation was used to fill in missing values for both parental income and education. Family Social Capital. Like others (Coleman 1988; Teachman et al. 1996), we conceptualize family social capital as having quantitative (e.g., family structure) and qualitative (e.g., parental interactions) components. Earlier research suggests that both of these components are associated with adolescent educational achievement (e.g., Hetherington 1998; Israel et al. 2001; Morgan and Sørensen 1999; Patterson et al. 1992; Tienda and Angel 1982). Family Structure and Size. Family structure is believed to affect well-being by influencing family functioning (McLanahan and Sandefur 1994; Thomson et al. 1994). For this reason a series of dummy variables were constructed based on the household roster. We use these to contrast youth who live with biological or adoptive parents (reference) with those residing in either a single parent or non-parent situation. Similarly, research on household composition suggests a link between household size and adolescent well-being, as adolescents in smaller households exhibit better educational achievement (e.g., Bridge et al. 1979, Nelson et al. 2001). Therefore a dummy variable that controls for household size is also incorporated. The reference group consists of households that contain no more than four members. Parent-Child Relationships. Both affective and instrumental ties between adolescents and their families and the parenting behaviors associated with them are viewed as a potential resources or forms of family social capital (Laosa 1982; Schneider and Coleman 1993; Smith et al. 1992; Stanton-Salazar 1997). Components of parent-child relationships, such as parents’ 4

To reduce the skewness of the original income variable in the Add Health Parents data set family income was ( Income + 1) 0.2 − 1 Income = 0.2 . For more on transformed using the Box-Cox family of transformations where Box-Cox transformations, see Box and Cox (1964).

Adolescent Educational Outcomes: Do Peer Networks Matter?

13

expectations for their children’s further education, or parental supervision and involvement have been documented to influence the educational outcomes of adolescents (e.g., Conger et al. 1994; Israel et al. 2001; Laosa 1982; Lee 1993; McLoyd 1998; McNeal 2001; Patterson et al. 1992). The index monitoring parents’ educational expectations was created from two items asked separately about mother’s and father’s expectations. Respondents were asked how disappointed each parent would be if they failed to graduate from college and high school. Responses range from low (1) to high disappointment (5). The reliability coefficient for these items is 0.82. Responses were averaged to create an index. Parental educational expectations capture cultural variation in the family’s emphasis on educational achievement, a family context characteristic that is often linked to immigrant academic success (Vernez et al. 1996). Parents’ involvement is constructed from nine items that inquire into the activities that parents and adolescents did together over the past four week period. For each parent adolescents were asked if they had done any of the following together: gone shopping, played a sport, attended a religious service or related event, talked about life, talked about a date or party attended, attended a movie, sports event, concert, play, or museum, talked about a personal problem, discussed grades or school work, worked on a school project, and talked about other school activities. Response choices are “yes” and “no.” The activities undertaken by the adolescent and at least one parent were then summed to form the index. The parents’ involvement scale has a Cronbach’s alpha of 0.72. Parental supervision is a variable that ranges from 0 to 3. It is constructed by summing affirmative responses to three items that indicate whether a parent is present in the home most or all of the time when the adolescent (1) goes to school in the morning, (2) comes home from school in the afternoon, and (3) goes to bed at night. The parental supervision scale has a Cronbach’s alpha of 0.68.

Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.

School-Level Variables The school-level variables examined monitor two fundamental aspects of each school’s student body: its racial/ethnic make-up and its socioeconomic composition. Although measures of school racial and ethnic composition (percentage Hispanic, percentage Asian, etc.) are not provided by the Add Health data, they can be directly calculated from the race/ethnicity responses of the student body. School-level race/ethnicity codes for these calculations are defined using the same codes earlier described for individuals. Note, however, that although we consider Asians as a distinct racial category in our analyses, they do not comprise part of our minority designation. Instead, our school-level measure monitoring the proportion of minority students in each school includes only Latinos and blacks. We do this because Asians are the nation’s most highly educated racial group and they attend the most integrated schools (Orfield and Yun 1999). Analyses presented in Table 4 further document the higher average academic achievement of Asian students vis-à-vis nonHispanic whites. To understand the importance of school-level SES this study employs the composite measure average SES, which was created by combining two school-level characteristics. More specifically, at the school-level the standardized scores for parental income and education were summed to create average SES. This is appropriate as these two variables are strongly intercorrelated at the school-level (Cronbach’s alpha = 0.90). Because both of these school-level SES measures are strongly

14

Igor Ryabov and Franklin Goza

skewed, they are transformed using the Box-Cox family of transformations. However, at the individual-level we consider it important for the present study to analyze these two indicators separately. This is done because some immigrant and minority groups experience status inconsistency, meaning that their educational attainment does not correspond with the occupations they occupy or the income they earn. This may explain why the reliability of the aggregate individual-level SES measure is lower (Cronbach's alpha = 0.71) than that of the school-level. According to Blau (1994), students cannot form friendships with students of other racial and ethnic groups if schools are homogenous. Interracial contact is a prerequisite for the formation of interracial friendships. For this reason, we included a measure of network segregation within schools. Many students named as friends are also members of the sample. This allows friends and their characteristics to be matched based on their survey responses. Both the Add Health inschool and in-home questionnaires asked students to list their five best male and female friends (including girlfriends and boyfriends). For each participating school the Add Health obtained a roster of its students and assigned them identification numbers. These rosters enabled students to find their friends in their school and a sister school. These identification numbers permit the direct determination of the race/ethnicity of adolescents’ friends. On the basis of friendship preferences, the Add Health constructed the modified Freeman’s race segregation index (1978). This school-level index is calculated as follows:

Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.

Segregation Index =

Expected

Ties

−

Expected

Observed

Ties

Ties

where ties refers to the total number of ties sent from a network member sharing the same race/ethnicity to all network members of other races or ethnic origins, summed across all race/ethnicity categories. The segregation index has a theoretical minimum of -1 (pure outgroup preference) and a theoretical maximum of 1 (pure in-group preference, or total segregation). A value of 0 indicates no group-preference, i.e. friendship ties are set randomly with respect to race/ethnicity. Table 2 presents the means for the dependent variables educational achievement and attainment when cross-tabulated with independent variables monitoring school-level SES, percentage of minority students, and the peer network segregation index. These continuous independent measures were divided into categories that correspond to high, medium, and low SES levels based on the 25th and 75th percentiles of their respective distributions. Table 2 demonstrates that at both times schools with the lowest SES and the highest percentage of minority students had the lowest GPAs. However, those attending schools with the highest level of peer network segregation had the highest GPAs. These results suggest a positive association between school SES and educational outcomes. Likewise, they also imply a positive association between peer network segregation and educational outcomes. However, the association between percent minority and educational outcomes is likely to be negative.

Adolescent Educational Outcomes: Do Peer Networks Matter?

15

Table 2. Average Educational Achievement and Attainment in Schools with High, Medium and Low School SES, Percentages of Minority Youth and Peer Network Segregation Index (N high school=80; N middle schools=49)

School-Level Variable School SES High Medium Low Percentage Minority in School High Medium Low Peer Network Segregation Index High Medium Low

Wave I Educational Achievement

Wave III Educational Attainment

2.98 2.79 2.66

14.03 13.10 12.38

2.66 2.80 2.87

12.49 13.08 13.39

2.92 2.86 2.65

13.51 13.12 12.04

Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.

Auxiliary analyses indicate that the school-level variables percentage minority and peer network segregation index are slightly skewed. Skewed variables can produce heteroscedasticity and inflated standard errors of regression estimates. These problems reduce the statistical power of significance tests and result in larger confidence intervals, which make the rejection of the null hypotheses more difficult (Stevens 1996). To remedy this potential problem we transformed the percentage of minority students and the racial segregation index using the Box-Cox family of log-linear transformations (Box and Cox 1964).5 Additionally, as a requirement of HLM, all school-level variables were school-mean centered.

Analytic Strategy The regression analyses presented below are performed using HLM. If we denote i as the ith student (level-1) and j as the jth school (level-2), the individual-level (level-1) model can be presented as follows:

Educationa l Outcome = β 0 j + β 1 j X 1ij + β 2 j X 2 ij + ... + β nj X nij + ρ ij , where β(0-n)j are regression coefficients of individual-level factors X(0-n)ij and ρij is normal error with mean 0 and variance σ2. The generalized formula of the school-level (level-2) intercept is:

5

We

used

the

Percentage Minority =

( Percentage Minority + 1) −0.2 − 1 − 0.2 ,

following formulas: and ( Segregation Index + 1) −2 − 1 Segregation Index = −2 . These transformations were obtained by running “BoxCox” macro from the SAS library.

16

Igor Ryabov and Franklin Goza

β 0 j = γ 0 + γ 1Y1 j + γ 2 Y 2 j + ... + γ n Y nj + ω 0 j where γ(0-l) are regression coefficients of school-level factors Y(0-n)j and ωj is the school-level error. Assuming no meso-level covariates, as we do in Tables 3 and 4, the regression coefficients β(0-n)j of individual-level factors are modeled as:

Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.

β 1 j = γ 1 + ω 1 j , β 2 j = γ 2 + ω 2 j and, in more general form: β nj = γ n + ω nj . When approaching the question of constructing 2-level HLM models, we used two modeling techniques available in HLM. The first is a linear model. It suits the interval/ratio character of the first dependent variable – academic achievement. Because the other dependent variable – academic attainment – has a dichotomous outcome, a Bernoulli model for binary data with logit link function was used to estimate the chance of graduation from high-school. One and possibly the most significant drawback of using a Bernoulli model is the inability to use sample weights. Therefore, Wave I grand sample weights were used to estimate achievement in Table 3, but not attainment in Table 4. Below we present three sets of analyses. In the first two we model educational achievement and attainment (see Tables 3 and 4, respectively) while examining the effects of individual-level controls, family structure and social capital, peer social capital, and school contextual characteristics. The third set of analyses (see Table 5) determines the effect of meso-level interaction terms on achievement and attainment. These terms monitor interactions between race/ethnicity (individual-level measures), mean school-level SES and peer network segregation (school-level measures). Parallel analyses are estimated for the achievement and attainment models. Table 3 presents results for all achievement models while Table 4 displays all attainment findings. Model 1 documents the effects of race/ethnicity and individual-level controls (age, gender, generational status, and frequency of involvement in extracurricular activities). Models 2, 3 and 4 add, respectively, family size and structure measures, parents’ SES effects, and family social capital variables. Together these three models test hypothesis V, where we expect that all types of family social capital will have a positive effect on student academic achievement. Model 5 adds the individual-level measure peer social capital as a way of testing hypothesis IV. Recall, we hypothesized a positive association between peer social capital and an individual student’s academic outcomes. To examine this hypothesis we treat peer network achievement as a proxy for peer social capital and monitor its relationship with student achievement and attainment. The final three models incorporate school-level factors, one at a time. Models 6, 7 and 8, respectively, add percentage of minority students, school SES and the peer network segregation index. These models test the hypotheses I, II and III, while controlling for the effect of individual-level factors.

RESULTS The first model of educational achievement consists primarily of individual-level demographic predictors (see Table 3). Model 1 demonstrates that many of these measures

Copyright © 2011. Nova Science Publishers, Incorporated. All rights reserved.

Adolescent Educational Outcomes: Do Peer Networks Matter?

17

were significant in the predicted directions. Blacks and Latinos are predicted to have significantly lower grades (p